cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 
Choose Language Hide Translation Bar
0 Kudos

Multivariate Contribution Plot Formatting Improvements

In JMP 15 it became possible to to view contribution plots for T2 within PLS and PCA models. To make these more usable, bring some functions from the [also new] Multivariate Drive Multivariate Control Chart to the PLS and PCA platforms, and add some new features.

 

Specifically:

  • Sort contribution plots by descending contribution, make sure it continues to be possible to scale the axis to only show the top few variables.
  • Transform the contribution plot so column labels are in rows to make them easier to read
  • Let the user display contributions for selected rows in a table (or mean for selected rows)
  • Link to uni-variate control charts for selected variables within the contribution plots or tables
  • Show contributions for T2 and DModX in both platforms, plus DModY in PLS
  • On both distance and T2 charts, let the user change the x axis to any other variable.  For example for time series data I frequently want this to be time.

 

As a use case, here is the chart I would use to explain to someone what is different about row 4893 of the Responses in the Probe sample data set. Right now this is difficult and resource intensive to draw. Please make this easy for new JMP users to create the same chart:

 

ih_0-1591814286181.png

2 Comments
Jeff_Perkinson
Community Manager
Status changed to: Yes, Stay Tuned!

JMP 16 will add an option "Model Driven Multivariate Control Chart for Saved X Scores" in each fit. It saves the x score formula first and then launch MDMCC launch dialog, which allows users to add/remove process, set historical or set time ID. Currently, only X scores are saved and processed.

Jeff_Perkinson
Community Manager
Status changed to: Delivered

With the release of JMP 16 you can now launch the Model Driven Multivariate Control Chart directly from the PLS platform.

2021-04-01_16-47-48.206.png